Injector transistors keep failing under heavy load

[Simscustoms]

Hi everyone, I'm building a 90 RX=7 FC with a 05 Miata engine in it for track use. The engine is the BP-Z3 VVT and I turbo charged it.
I'm using a MS-2 on a V3.0 B&G board that I put together. I'm also using JBPerf dual VR board for my tach inputs and the
4-channel board for logic level ignition outputs and injector outputs. I added a PWM output for a 2 wired idle valve,
a boost controler out wich is not being used and is deactivated at the moment and I also added a knock sensor input coming from
a knocksenseMS that I have setup on the car so far.I havn't used the knock input as a safety yet. The VVT is being controlled by VVTunner from DIYautotune and the tach signal is going in/out as pass throu to the ms2. I got all the sensors calibrated and got
the car starting, idling and driving great after some tuning and adjustments until I went WOT as soon as I let off and the AFR droped to full rich I knew something was wrong. After towing it back to my shop and and a bit of tersting it was obvious I had a transistor stuck to ground.

Here is how I put together the inejctor output circuits with a VND7N04TR-E transistor and a 1k and 100k resistors as foudn in the instructions.
This circuit is pretty much what the ms3x has but the ms3x uses VND5N07. The drivers each drive 1 RC SL-9 550 CC injector.

So here is what I tried:
-I replaced the transistor and it fixed it until I went WOT again and had 3 of them fail at the same time
-Inspected all the connectors and wiring harness
-set the proper injector dead time as per the manufacturer's chart
-Increased the input wires from the v3.0 board to the 4 channel board from 26 to 22 gauge and also made sure the DB15 3 inejctor grounds were adequate.
-researched how to properly solder surface mount transistors and resoldered all 4 new transistors.
The car started nice and ran nice until WOT again had I think 2 transistor fail that time.
-Replaced the failed transistors and built a test bench for the inejctors to squirt in cups instead of the engine. The bench test uses the cars wiring and the jimstim is wired to the power, ground and fp output turns on the fp, ign, inj relays. I could not make the injector outputs fail this way until I hooked up a runing car to the test car's battery to increase the voltage and then I had 2 drivers fail.
-The first board I was testing with had bin throu a lotover the past 12 years so I decided to put together a new MS-2 V3.0 board with a new
4 channel board and new components and Then tested the new ecu up to 6000 rpm and 20 ms pulse width and everything would stay nice and cooled
measured with my laser thermometer. even with a car pumping 13.5 volts into the battery for the bench test. Then I got the car all back together and went for a test drive and had the same thing happen again. Got it running good until I hit WOT and the injector transistors fail.
This time 2 more transistors failed.

Things I noticed
when at full boost the PW is around 16 MS wich is lower than what I tested at on the bench. and on the road it only goes up to it for half a second.
I had them running at 20 ms and 5-6000 rpms on the bench for probrably 30 sec. straight multiple times and they would stay at room temp.

I was thinking of trying VND14NV04 transistors next.
I was asking my self also Could the Injectors be to blame for this?

2018-08-18_01.43.59.msq

2018-08-19_19.22.04.msl

[Marek]

The transistors are just a switching device. The amount of current pulled by the circuit depends on how low the impedance of the circuit is when it is connected.

The MS3X circuit as described expects to operate high impedance injectors which do not pull a substantial current.

When a high current usage injector (i.e. a low impedance injector) needs to be switched on, it is done via the mechanism of deliberately altering the current usage profile:- a high current is used to open the injector, but as soon as it opens, the current is throttles back to the minimum needed to keep it it open and flowing fuel. This is done either wiith hardware -a dedicated peak and hold driver - or via software using current limiting via pulse width modulation.

If you ask the transistors to switch and flow too much current for too long, they will overheat, as you have observed.

kind regards
Marek

[Simscustoms]

Thanks Marek, but as far as I know the RC SL-9 are high impedance injectors. I have measured the resistance of each injector and they all measure at 13 ohms. Is there something I'm missing?

Sent from my SM-G900W8 using Tapatalk

[DaveEFI]

Thanks Marek, but as far as I know the RC SL-9 are high impedance injectors. I have measured the resistance of each injector and they all measure at 13 ohms. Is there something I'm missing?

Sent from my SM-G900W8 using Tapatalk

13 ohms would draw roughly 1 amp per injector. The VND7N04TR-E is an integrated circuit with overload etc protection built in (IIRC), and should be quite happy with a 1 amp load - and quite a bit more.
So not quite sure what is going on here.

[LAV1000]

Those VND7N04TR-E do you trigger them with 5 or 12Vdc?

[Simscustoms]

Im really stumped too. I did some bench testing and had them runing at a higher load then they see at wot for a mich longer period and they stayed at room temp. But as soon as I hit full boost with the car I will have atleast one that will fail instantly without any visual signs. I have a 5 amp fuse for the voltage to the injectors and thought it would blow befor the transistor do but even the fise wasnt even hot after the last transistors failed. I checked the temps on the injector board right after the last 2 failed and the temp was around 175F.

Sent from my SM-G900W8 using Tapatalk

[Simscustoms]

Those VND7N04TR-E do you trigger them with 5 or 12Vdc?

I have them wired as per JBPerf. instructions found here.
http://jbperf.com/quad_ign_inj/index.html
I am using the 4 channel board version 1.1 with thr logic level ignition outputs wich seem to work great with my toyota cop. I can with it on the jim stim and measure the voltage from the trigger to the transistors to confirm the voltage.

What should the trigger be at?

Sent from my SM-G900W8 using Tapatalk

[LAV1000]

If you trigger them with 5Vdc use 330 Ohm resistors.
If you trigger them with 12Vdc use 1 kOhm resistors.

When using the 1 kOhm at 5Vdc it takes longer to switch fully open, so there is more heat build up in the unit.
If you need to order new ones take the later type which is used ms3*** they are a little better at over voltage protection.

And check grounds, over voltage protection need a a solid ground to work properly.

Side note, alternator/rectifier might also cause problems.
When alternator has a big load and load drops voltage may overshoot, this is behavior of a bad alternator/rectifier.

[racingmini_mtl]

When using the 1 kOhm at 5Vdc it takes longer to switch fully open, so there is more heat build up in the unit.
If you need to order new ones take the later type which is used ms3*** they are a little better at over voltage protection.

While switching will be slower with a 1k than with a 330R resistor, I haven't heard that being an issue. But that can't hurt to change. And which part number is now used in the MS3? The ones mentioned by the OP is more robust then the original VND5N07.

Jean

[LAV1000]

When using the 1 kOhm at 5Vdc it takes longer to switch fully open, so there is more heat build up in the unit.
If you need to order new ones take the later type which is used ms3*** they are a little better at over voltage protection.

While switching will be slower with a 1k than with a 330R resistor, I haven't heard that being an issue. But that can't hurt to change. And which part number is now used in the MS3? The ones mentioned by the OP is more robust then the original VND5N07.

Jean

Nominal current might be bigger but Drain to Source clampvoltage is 42V and on the VND5NO7 it is 70V.

Like I also suggested check grounds and alternator/rectifier.

Jean, I know you make a lot of good and wel tested stuff.
So there might be something wrong in how OP did put the hardware together.

[racingmini_mtl]

When using the 1 kOhm at 5Vdc it takes longer to switch fully open, so there is more heat build up in the unit.
If you need to order new ones take the later type which is used ms3*** they are a little better at over voltage protection.

While switching will be slower with a 1k than with a 330R resistor, I haven't heard that being an issue. But that can't hurt to change. And which part number is now used in the MS3? The ones mentioned by the OP is more robust then the original VND5N07.

Jean

Nominal current might be bigger but Drain to Source clampvoltage is 42V and on the VND5NO7 it is 70V.

Like I also suggested check grounds and alternator/rectifier.

Jean, I know you make a lot of good and wel tested stuff.
So there might be something wrong in how OP did put the hardware together.

We've had some discussion on this and I can't see any issue at least from description and pictures but I haven't had the board in hand nor do I know all the details of the complete setup.

And the lower clamping voltage should make the setup more robust by decreasing the flyback power. It will have an impact on closing time but that should not be significant. The most important things are the lower RDSon and the higher avalanche energy rating. The VND14NV04 has a much lower RDSon.

Jean

[Simscustoms]

I am unable to measure the voltage of the trigger with my multi meter. I am using a ms2. Can one of yous please tell me what's the voltage of the injector triggers from pin 4,5,39 and 40 on the ms2?

I wasn't sure either to use a 330 or 1k ohm so I looked up the ms3 diagram and decided to use the same. I didn't know or think the trigger voltage could be different from ms2 to ms3.

I appreciate all the comments and am really glad to have everyone helping me out so far.

Yes, I agree JBPerf's boards are very well designed and convenient, they clean up the ecu very well. I don't intend on bashing any of his boards. The problem is probrably me that has something wrong somewhere. I thought maybe there was a problem with the power circuit so now that I have a new v3.0 board and ms2 processor, that should eliminate that possibility. I successfuly ran and tuned half a douzaine mazda B6 or BP engines with the first ms2 v3.0 I had but I ran them with distributors or different tach signal inputs from the older models. This is the lastest BP engine from an 05 Miata and I had to do a lot of research and learning to get to this point. There is a lot of misleading information online that could of thrown me off. I've trying to stick to the msextra manual as much as possible. The toughest part was the tach signal. I tried to reproduce what I found in the manual for the cam input and couldn't get it to work. The JBPerf dual VR board took care of that no problem.

Here's a few more details on my ECU. Here's a photo of each side of the new board. You will notice I have some of the original fuel injector circuit populated on the v3.0 board. I installed the circuit to control a boost control solenoid as follows in the last photo. I do not have it activated. You will also notice on the bottom of the board in the proto area the pull up resistors and input and output wires for the dual vr board wich takes care of the crank/cam signals. right next to the dual vr circuit you will see a zenner diode and a resistor that compose the knocksensems input as seen on step 3.5.7 of the ms2 msextra manual. You will also notice that the board was still full of resin and a bit of solder residue in these pics. I took them right after completing the ecu last Friday. I have since gave it a good cleaning befor starting the testing(last saterday) and frying the transistors again sunday.

front of v3.0 board.jpg

back of v3.0 board.jpg

boost controler.jpg

[dontz125]

The VND14NV04 has a much lower RDSon.

Looking at both DS just now, some of the RDSon graphs are almost identical; a few are different by 20-30 mA. I don't see where that would cause critical failures of one vs the other. What I *did* notice was the rather draconian power derating curve vs temperature, completely linear from 100% at 25C to 0% at 150C. Also, the storage temperature is -55 to 150C, but the minimum shutdown temp is also 150C; this implies (to my mind) that the shut-down on a particular device could be *higher* than the recommended safe storage temperature. There is no value given for the max junction temperature ("Internally limited"), but I've seen a lot of 160-175C values in similar devices.

If these devices have thermal limiters that don't necessarily kick in until the termperature has reached dangerous or damaging levels, it could explain why some VND chips live while others die a miserable death.

[racingmini_mtl]

Looking at both DS just now, some of the RDSon graphs are almost identical; a few are different by 20-30 mA.

VND5N07 is rated at 0.2 Ohm, VND7N04 is rated at 0.14 Ohm and VND14NV04 is rated at 35mOhm. Of course that's not true for all conditions but that's a major difference in heat generated in the driver while the injector is on.

Jean

[Simscustoms]

quad_ign_inj_v1_1.png

I did a bit of research and found that the ms2 injector outputs are 5 volts so then I should use 330 ohm reistors for R5 to R8?
And R9 to R12 should be 100 K Ohm? right?

If I use a VND14NV04, does that change the value of the resistor I should be using?

I had 1 k ohm resistors in the place of R5 to R8, Is that what was causing my problem?

Thanks again


Simon

[racingmini_mtl]

1k Ohm is what has been used in the MS products (at least for the ones I know) and what I have recommended and provided in my kits. Using 330 Ohm resistors would work and in theory would allow faster switching and potentially less heat. At worst, it will do nothing and at best it will make enough of a difference to help in your case. Using the VND14NV04 won't change that and you still want to use 100k resistors for R9-12

I wish I could provide you with a more definitive answer but I don't see what is different in your setup that creates the problems you see. So using 330 Ohms resistors and VND14NV04 drivers is likely the best option for you at this point.

Another option that you may want to explore is to use TO220 drivers which will be bigger and have better heat dissipation due to their physical format and associated heat sink. But that means either using the MS board heat sink for the drivers or drilling the case and using it as a heats sink for the driver. I have made a new board that uses the same footprint as the ignition IGBTs on the board you have but for TO220 injection drivers; the board also has the logic level ignition drivers and it uses a DB9 connector. You can contact me directly if you want to try this option.

Jean

[dontz125]

Just talking out my, ummm, ear; the data sheeet gives 550uA max input current with 10v input. I'm not entirely sure a series resistor is needed at all - the input pin isn't actually connected to the MosFET gate.

I don't know where they get the 'rated' RDSon, but the various graphs are very similar, given the same Ta, Vds and Vin.

[Simscustoms]

Yes, I don't understand what's so special with my setup. I thought for sure this had bin done thousands of time befor . I have some 330 ohm ressitors I will swap out for the 1k ohm and I should be receiving the VND14NV04 tomorrow. I will put this together tomorrow hopfully and more testing.

[racingmini_mtl]

Just talking out my, ummm, ear; the data sheeet gives 550uA max input current with 10v input. I'm not entirely sure a series resistor is needed at all - the input pin isn't actually connected to the MosFET gate.

At least one of the datasheets says that the minimum resistance is 10 Ohms.

[jsmcortina]

I'm wondering if something other than the current capacity is the problem.

Depending on where the main ground is taken, there could be a ground offset that takes the gate negative. That could break the device. It seems unlikely that the gate could get too positive.
Consider adding a 1N4148 diode from ground to gate to protect against negative voltages there.

James